Playwright Verification
Playwright-based UI verification — screenshot evidence, navigation testing, accessibility checks
Tags
Overview
Purpose
Playwright-based UI verification — screenshot evidence, navigation testing, accessibility checks
Rules
PLW-001: LLM MUST use Playwright to verify work appears in the application UI before claiming done.
LLM agents cannot see rendered UI without browser automation. Playwright bridges the gap between code changes and visual verification.
Verification: Playwright session executed before any UI completion claim.
PLW-002: Take screenshot after verification. Store in task artifacts as evidence.
Screenshots are immutable evidence of state at verification time. They enable asynchronous review without re-running verification.
Verification: Screenshot file exists in task artifacts directory.
PLW-003: Verify navigation to new content works: click link, verify content loads.
Routing bugs are invisible in component-level testing. End-to-end navigation proves the full path works.
Verification: Playwright test: navigate to page, assert content loaded.
PLW-004: Verify data displayed matches expected values from fixtures or source.
Rendering correct structure with wrong data is a silent bug. Data verification catches transform and binding errors.
Verification: Playwright test: assert text content matches fixture values.
PLW-005: Test error states: missing data, invalid input, network failure.
Happy-path-only testing misses the states users encounter most. Error states need explicit verification.
Verification: Playwright test: error scenarios tested with assertions on error UI.
PLW-006: Run accessibility checks on new UI components via Playwright.
Accessibility violations are invisible to sighted testers. Automated checks catch missing ARIA labels, contrast issues, and keyboard navigation gaps.
Verification: Playwright accessibility audit: no critical violations on new components.